which is cheaper (cost involved in both fabrication and testing) a single board with 9 DSP's or 3 boards with 3 DSP's each with proper sockets to communicate. amar
single board vs multiple board
Started by ●May 2, 2004
Reply by ●May 2, 20042004-05-02
amara vati wrote:> which is cheaper (cost involved in both fabrication and testing) > > a single board with 9 DSP's or 3 boards with 3 DSP's each with proper > sockets to communicate. > > amarIs this a joke? Edging and handling three small boards clearly costs more than one, even if the total board areas are equal. Test points from which signals can be drawn for analysis cost less than interconnection sockets and don't imply a reliability risk. Did you think it could be otherwise? Did you think? (I'm grumpy this morning.) Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 2, 20042004-05-02
Jerry Avins wrote:> amara vati wrote: > >> which is cheaper (cost involved in both fabrication and testing) >> >> a single board with 9 DSP's or 3 boards with 3 DSP's each with proper >> sockets to communicate. >> >> amar > > > Is this a joke? Edging and handling three small boards clearly costs > more than one, even if the total board areas are equal. Test points from > which signals can be drawn for analysis cost less than interconnection > sockets and don't imply a reliability risk. Did you think it could be > otherwise? Did you think? (I'm grumpy this morning.) > > JerryI would tend to say that original question was poorly defined. As you point out, handling 3*M boards costs more than handling M boards. But does that not implicitly assume a *ZERO* failure rate in QC? Might not economies of scale also come into play? What if there was a market for 1000 9DSP boards and a market for 4000 3DSP boards? The physical environment was also unspecified. What if this undefined application card assumed installation in a standard PC and BNC connectors were required? I can see 3 BNC's on a standard card. 9 requires "gyrations". All to say original question was poorly defined.
Reply by ●May 2, 20042004-05-02
Jerry Avins wrote:> amara vati wrote: > >> which is cheaper (cost involved in both fabrication and testing) >> >> a single board with 9 DSP's or 3 boards with 3 DSP's each with proper >> sockets to communicate. >> >> amar > > > Is this a joke? Edging and handling three small boards clearly costs > more than one, even if the total board areas are equal. Test points from > which signals can be drawn for analysis cost less than interconnection > sockets and don't imply a reliability risk. Did you think it could be > otherwise? Did you think? (I'm grumpy this morning.) > > JerryBoy, you _are_. I can see where the 3-identical-board idea may be more cost effective, particularly if the solution is low volume production or needs to be scalable, but the real answer depends on way too many variables to know. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●May 2, 20042004-05-02
Jerry Avins wrote: (snip)> Is this a joke? Edging and handling three small boards clearly costs > more than one, even if the total board areas are equal. Test points from > which signals can be drawn for analysis cost less than interconnection > sockets and don't imply a reliability risk. Did you think it could be > otherwise? Did you think? (I'm grumpy this morning.)Price Pentium4 motherboards (I forget the socket number by now), in single, double, and quad versions. I am pretty sure that the prices won't follow your rule. That said, I have no idea how prices for DSP boards go. -- glen
Reply by ●May 2, 20042004-05-02
glen herrmannsfeldt wrote:> Jerry Avins wrote: > > (snip) > >> Is this a joke? Edging and handling three small boards clearly costs >> more than one, even if the total board areas are equal. Test points from >> which signals can be drawn for analysis cost less than interconnection >> sockets and don't imply a reliability risk. Did you think it could be >> otherwise? Did you think? (I'm grumpy this morning.) > > > Price Pentium4 motherboards (I forget the socket number by now), > in single, double, and quad versions. I am pretty sure that > the prices won't follow your rule. > > That said, I have no idea how prices for DSP boards go. > > -- glenI wasn't thinking identical, but rather a matched heterogeneous set. My gaffe highlights the worth of the good advice to say nothing when one has nothing good to say. Following it is not only more polite, but likely to avoid embarrassment. Amara, I apologize. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 3, 20042004-05-03
Hi, sorry I wasnt descriptive enough in posting my query. I am actually talking about three perfectly identical boards. so as far as testing in concerned, I would have to test only a single 3-DSP board. Moreover, I am looking for scalablility. Like I would club as many boards as the requirements dictate. but what about the mass production costs. I feel that should also come down. thanks for your suggestions. amar Jerry Avins <jya@ieee.org> wrote in message news:<4095a85c$0$3044$61fed72c@news.rcn.com>...> glen herrmannsfeldt wrote: > > > Jerry Avins wrote: > > > > (snip) > > > >> Is this a joke? Edging and handling three small boards clearly costs > >> more than one, even if the total board areas are equal. Test points from > >> which signals can be drawn for analysis cost less than interconnection > >> sockets and don't imply a reliability risk. Did you think it could be > >> otherwise? Did you think? (I'm grumpy this morning.) > > > > > > Price Pentium4 motherboards (I forget the socket number by now), > > in single, double, and quad versions. I am pretty sure that > > the prices won't follow your rule. > > > > That said, I have no idea how prices for DSP boards go. > > > > -- glen > > I wasn't thinking identical, but rather a matched heterogeneous set. My > gaffe highlights the worth of the good advice to say nothing when one > has nothing good to say. Following it is not only more polite, but > likely to avoid embarrassment. Amara, I apologize. > > Jerry